long-term transportation electricity use: estimates and ...€¦ · executive summary please see ...
TRANSCRIPT
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
i Electric_Transport_Review_Copy _10_1_17
Long-Term
Transportation Electricity Use:
Estimates and Policy Observations
By
Peter Fox-Penner
Will Gorman@
Jennifer Hatch#
Boston University
Institute for Sustainable Energy
595 Commonwealth Ave
Boston, MA 02215
Preface This is a very long and somewhat detailed literature review and discussion. Its purpose is to stimulate
further discussion, research, and policy formulation and give the power industry some idea of what
might be coming in the way of EV sales growth. Due to the paper’s length, most readers will want to
read only the executive summary or selected parts. Depending on reviewer feedback, the paper may
remain in its present form, become a book, or be divided into smaller articles for review at journals.
Professor of the Practice, Questrom School of Business at Boston University; Director, Boston University Institute for Sustainable Energy, [email protected] @Ph.D. Program, Energy and Resources Group, University of California, Berkley, Ca. #Research Fellow, Boston University Institute for Sustainable Energy, [email protected] – Corresponding Author
PRELIMINARY DRAFT
COMMENTS WELCOME
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
ii Electric_Transport_Review_Copy _10_1_17
Abstract
In this paper we model three layers of transportation disruption – first electrification, then autonomy,
and finally sharing and pooling – in order to project transportation electricity demand to 2050. In
addition, we consider three “wild cards” that have the potential to influence LDV travel in especially
unpredictable ways. Using an expanded kaya identity framework, we model vehicle stock, energy
intensity, and vehicle miles traveled, progressively considering the effects of each of these three
disruptions. We find that energy use from light duty vehicle (LDV) transport will likely be in the 570 TWh
to 1140 TWh range, 13% to 26%, respectively, of total electricity demand in 2050.
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
iii Electric_Transport_Review_Copy _10_1_17
Acknowledgements This paper would not be possible without the extensive input, advice, and deep expertise of several
thought leaders in the realm of electric and automated vehicles and transit futures. In alphabetical
order, the authors would like to thank:
Heidi Bishop, Austin Brown, Kristopher Carter, Christos Cassandras, Stacey Davis, John Dulac, Reid Ewing, Ryan Falconer, Alison Felix, Tyler Folsom, Lewis Fulton, Marianne Gray, Jeff Gonder, John Helveston, Robert Herendeen, Alejandro Henao, Ryan Hledik, Josh Johnson, Henry Kelly, Kyeil Kim, Kara Kockelman, Ken Laberteux, Ryan Laemel, Amory Lovins, Andy Lubershane, Roger Lueken, Ryan Rucks, Susan Shaheen, James Schulte, Dan Sperling, Joshua Sperling, Tom Stephens, Jerry Weiland, Jurgen Weiss, and Johanna Zmud for their guidance in writing various segments of this paper. We would also like to extend an enormous thanks to Bai Li and Xuehui Sophia Xiong for their research assistance. In addition to general assistance, Xuehui Sophia Xiong examined emissions factors and Bai Li prepared the EV energy intensity projections through 2050. We gratefully acknowledge financial support from the Hewlett Foundation, the Energy Foundation, and Boston University.
Author Disclosures In addition to his Boston University duties, Peter Fox-Penner serves as chief strategy officer for Energy
Impact Partners, which owns interests in storage and EV infrastructure firms, among others. In addition,
he is on the Advisory Board of EOS Energy Storage. Jennifer Hatch and Will Gorman have no financial
interests in the transport or power sectors.
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
4 Electric_Transport_Draft _10_1_17
List of Acronyms, Abbreviations, and Initializations AEV Autonomous electric vehicle AET Autonomous electric taxis AV Autonomous vehicle BEV Battery-Electric vehicle BSUV Battery SUV CSUV Conventional SUV CV Conventional car DRS Dynamic Ridesharing Services EI Electric intensity (kWh/mile) EAV Electric autonomous vehicle EHV Electric hybrid vehicle EV Electric vehicle GDP Gross domestic product GHG Greenhouse gas GWh Gigawatt hour HDV Heavy-duty vehicle HEV Hybrid electric vehicle ICE Internal combustion engine kWh kilowatt-hour LDV Light-duty vehicle MPG Miles per gallon PHEV Plug-in hybrid electric vehicle PHSUV Plug-in hybrid SUV PM Passenger-mile SAV Shared autonomous vehicle SBEV Shared battery electric vehicle SBSUV Shared battery SUV SPHEV Shared plug-in hybrid electric vehicle SPHSUV Shared plug-in hybrid SUV TWh Terawatt hours WT/mile watt hour/mile UI use intensity, same as EI when referring to electric vehicles VMT vehicle miles traveled
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
1 Electric_Transport_Draft _10_1_17
Executive Summary
The transportation sector is now facing trifecta disruptions of electrification, sharing, and autonomy –
disruptions known in some transport circles as the “Three Revolutions.” Together, these disruptions are
expected to have profound impacts across developed world economies, from the auto industry, to the
labor force, to family lifestyles and more. In an attempt to explore a small corner of impact from these
revolutions, this paper attempts to quantify the electricity needed to power light duty passenger electric
fleets.
In contrast to many other works on the subject, our focus is on the aggregate national electrical energy
needed to power the light-duty vehicle fleet between now and 2050. From the standpoint of climate
policy the electric energy we need is the single most important indicator of our need for carbon free
power. As we stick to this specific focus we do not analyze or discuss many other important aspects of
the growth of electric transport on electric utilities, including charging patterns or methods, integrating
EV demand response, or energy sourcing. We also do not produce new forecasts of EV and AV sales or
the underlying costs of owning and operating vehicles over time.
We review and rely on several industry forecasts to create our own EV scenarios. Our work can be
viewed as attempting to improve upon, or at least add usefully to, the handful of studies that examine
long-term transport power demand. The studies we reference and rely upon can be referenced in part
II.A – Prior Work.
Research Approach
Transport energy and emissions are often forecasted by (1) estimating the vehicle-miles that will be
traveled (VMT) , using well-established models benchmarked from prior changes in travel on these modes
over decades; and (2) multiplying VMT times the energy use per vehicle-mile which can be forecasted by
analyzing current efficiencies, fuel economy rules, fleet composition shifts, and similar factors. To address
difficulties in depending upon aggregate VMT forecasts, in our work we use a conceptual framework
based on an expanded kaya identity, and then apply the framework in the three “layers” and further
adjustments as explained below. The disaggregated kaya identity we use is:
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
2 Electric_Transport_Draft _10_1_17
Where the stock in year t of EVs of a motorized vehicle type i is denoted by ĸi,t; vi,t is the average miles
travelled by that vehicle type in year t; and eni,t is the average power use of the vehicle type i per mile
travelled during year t.
One obvious deficiency with our approach is that VMT, mode share, and electric intensity are all
interdependent. Every decision to take a trip is a function of the underlying drivers of travel and the
mode choices for each chosen trip – time cost, money cost, and other costs and benefits for each mode
option. This interdependence operates differently in the short-run and the long-run. In the short run,
the choice of travel mode is approximately fixed by the state of technology, existing infrastructure and
vehicle stocks, and current arrangements such as transit schedules and the accessibility of EV chargers.
In the long-run, every one of these trip choice determinants changes in a path-dependent manner.
Moreover, in the long-run the choice is nested, first in a choice of own/share a vehicle by type and then
whether to use that vehicle for a trip.
We break through this deep interdependence using a very simple and inelegant approach. We first
posit a baseline in which none of the Three Revolutions occurs. In this baseline scenario, we generally
adopt the view of Litman, Circella, et al, and the Federal Highway Administration that per-capita LDV
travel by Americans has hit its peak and is likely to decline, but for the potential effects of the Three
Revolutions. The effect of the Three Revolutions is then factored into our implicit baseline in three
“layers” of calculations.
The first layer is electrification, an interim scenario in which the only major change is the availability of
EVs as an alternative to CVs, i.e. without changes in ownership models, autonomy, shared modes, or any
urban design changes not already embedded in conventional forecasts. The next layer of our
calculation modifies this interim case to reflect the onset of autonomous vehicles. As many researchers
are predicting, AVs will have many complex effects on travel demand, amounting on net to a significant
and perhaps very large increase in VMT. Conversely, AVs will reportedly enable savings in EI through
network and vehicle management approaches not available to conventional vehicles, and (much later,
we think) energy-reducing changes to the vehicles themselves. In this layer, we first survey AV
penetration predictions and adopt an AV penetration base case and a second, more aggressive
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
3 Electric_Transport_Draft _10_1_17
sensitivity case. In the final layer, we add the potential impacts of the new pooled and shared modes,
including integrated multimodal systems, also called mobility-as-a-service, among other ideas. The
figure below illustrates our conceptual approach.
Figure 1: Conceptual Approach
Conventional Ownership
For our analysis, we begin with forecasts from industry groups that have projected electric vehicle sales,
with or without visible adjustments for the growth of autonomous driving or new ownership models.
After examining several commercial projections, we assume that EV adoption follows a similar
technology adoption curve described by Everett Rogers’s Diffusion of Innovations theory.
Electric vehicle sales, though, do not represent the actual stock of electric vehicles in a given year that
would consume electricity. Rather, electricity consumption would be driven by the total number of
vehicles on the road, which is affected by car retirements as well as sales. To inform our estimate of
electric vehicle stock, we rely on the survival rates for conventional cars and light trucks provided by the
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
4 Electric_Transport_Draft _10_1_17
Oak Ridge National Laboratory’s Transportation Energy Data Book (Oad Ridge National Lab, 2016).
Details of our assumptions and estimations can be found in section III.A – EV Projections Under
Conventional Ownership.
The figure below shows the stock of electric LDVs in our high and low cases.
Figure 2: Electric Vehicle Stock Under Conventional Ownership
\
We next estimate how many miles each conventionally-owned vehicle in our stock will drive annually,
using 2015 Idaho National Labs survey data for BEVs and PHEVs. The eVMT values for both PHEVs and
BEVs, though, are noticeably lower than the average VMTs of ICE LDVs today. For ICE cars in 2015, the
average annual VMT was 11,327 miles and for SUVs it was 11,855 miles. Current models of electric
vehicles often do not have the same drive range as the ICE equivalent vehicle due to the limitations of
current battery technology. However, we expect the annual miles driven using electricity to increase as
battery technology continues to improve and battery ranges increase. In order to capture the effect of
battery improvement for the future years of our analysis, we fit a curve to projected battery energy
density increases and use the percent increase over time to gross up the total electric vehicle miles for
both PHEVs and BEVs.
We then use the annual VMT for ICE vehicles as the baseline distance the average electric car owner
would drive in a year under conventional ownership prior to autonomy. In other words, in this scenario
we linearly trend annual eVMTs for PHEVs and BEVs from their current average levels to the average
VMT of conventional vehicles in 2015 as reported by the FHWA. Details of our methodology can be
found in section III.B – VMT Assumptions.
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
5 Electric_Transport_Draft _10_1_17
Even if we assume the ownership model will not change, as we assume for the base layer, EVs are likely
to become steadily more energy-efficient over time. We therefore estimate trends in energy efficiency
improvements based on well-established theories of the returns to R&D and manufacturing learning
curves.
The combination of these assumptions represents the three main inputs to the kaya identity
formulation. Our EV projections, eVMT estimates, and expected vehicle energy intensities are multiplied
by each other to calculate our conventional ownership base case (or stage one) electricity consumption
projections, shown in the table below:
Table 3: Conventional Ownership Results
Overall, we project 2050 electricity demand of 890 TWh and 510 TWh in our high and low cases,
respectively. These figures represent roughly 23 and 13% of the current electricity demand of 3900 TWh
and 20 and 11%, respectively, of EIA’s projected 2050 electricity consumption of 4,500 TWh.
Case Year
Total
Number of
EV in Service
Portion
Stock
Electric
Total
Number
of AV in
Service
Fleet
Average
eVMT /
Vehicle
Fleet
Average
Efficiency
Total
TWh
Total
TWh EV
Bump
(%) (per yr) (kWh/mile) (TWh) (TWh)
2015 406,076 0.2% 0 7,179 0.32 0.9 0.9
2025 17,086,996 6.6% 0 10,075 0.34 59.0 59.0
2030 52,378,548 19.7% 0 10,734 0.33 187.9 194.0
2040 166,919,164 59.6% 0 11,039 0.32 593.0 651.9
2050 251,742,035 85.4% 0 11,231 0.31 886.2 973.8
2015 406,076 0.2% 0 7,179 0.32 0.9 0.9
2025 7,063,273 2.7% 0 10,061 0.34 24.3 24.3
2030 20,532,231 7.7% 0 10,729 0.33 73.6 76.0
2040 81,511,381 29.1% 0 11,049 0.32 289.2 317.9
2050 145,941,420 49.5% 0 11,236 0.31 511.7 562.3
Base Low
Base High
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
6 Electric_Transport_Draft _10_1_17
Impact of Autonomous Vehicles
Next, we examine the power impacts of commercially available, fully-self-driving (“autonomous”) light-
duty vehicles. We oversimplify by treating the transition to AVs as a bright line before and after Level 4
or 5 AVs sold and allowed to be used with relatively few restrictions. Our projections show the national
totals, increasing as the number of areas and vehicles sold both rise.
There is a cacophony of opinions as to when and how the autonomy revolution will occur – not to
mention its implications for travel, the economy, and our built environment. On one end stand highly
optimistic writers such as Aribib and Seba, who predict that AVs will handle 95% of all passenger-miles
by 2030, all but ending individual auto ownership. At the other extreme, researchers such as Litman
(Litman, 2017) and Nieuwenhuijsen (Niewenhuijsen, 2015) predict that 100% level 5 autonomy in the
fleet will not occur until 2070 or later. Beyond differences in numerical outcomes, some of these
estimates come with somewhat concrete scenarios or narratives as to how the AV market will unfold
with respect to regulatory approval, cost reduction, consumer choice shifts, and urban infrastructure
changes.
Amongst all these considerations the work we find most convincing is Lavasani, Jin, and Du’s (Lavasani,
Jin, & Du, 2016) estimates of Bass or “S-curves” using parameters selected by comparing AVs to other
types of technologies, similar and dissimilar, for which there are full adoption histories. The results of LJD’s
base estimate, is that cumulative AV sales rise from 1.3 MM in 2030, five years after introduction, to 70
MM by 2045 and saturation (i.e. no further growth in AV sales) by 2060. We also create estimates of
electricity use for the A&S scenario, which we consider a highly aggressive upper bound on AV use. If
nothing else, this allows us to estimate a range of possible outcomes.
There is widespread agreement that vehicle autonomy will trigger significant changes in the travel
patterns of many Americans (along with changes in EI, explored later). Some of these changes will
reduce VMT, while others are expected to increase it significantly. These effects include increased road
capacity as AVs travel smoothly at close intervals, lower time cost for drivers as driving time is freed for
leisure or work activities, and increased access as children, the elderly, and disabled use autonomy to
increase mobility – the results of these effects are outlined in the table below.
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
7 Electric_Transport_Draft _10_1_17
Table 4: VMT Effect of Automated Vehicles
AVs similarly have significant effects on the energy intensity used per mile of any given vehicle type.
These effects include: traffic smoothing due to their ability to immediately see and respond to traffic
conditions; better intersection management to reduce starts and stops and therefore reduce energy
use; faster travel which will increase EI as AVs travel at higher speeds; platooning, which will reduce air
resistance; and rightsizing, where smaller and lighter vehicles will be available because safety features of
conventional vehicles will no longer be necessary. The net EI of AVs is summarized in the table below:
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
8 Electric_Transport_Draft _10_1_17
Table 5: EI Effect of Automated Vehicles
Effect Impact Timing
Traffic Smoothing -15% 50% reduction in technology
improvements in EI for the first
10 years, then linear phase- in
from 2035
Intersection Management -4% Linear phase-in for urban EVs
starting in 2035 and fully
implemented by 2055
Higher Average Speed +8% Linear phase-in from 2030-2035
Platooning -2.5% Linear phase-in from 2030-2035
Rightsizing/Weight Reduction -50% Phased in linearly at 1% per year
or 1.5% per year starting in 2040
Pooling, sharing, and seamless
We next add a third layer to our electric power forecasts, the impacts of many shared and pooled modes
and businesses including various forms of what are being called “mobility networks.” We examine
literature surrounding non-pooled dynamic ridesourcing, traditional carpooling, car-sharing, and pooled
dynamic ridesharing or ridesplitting, and in the end conclude that these three phenomena, while
extremely important for the ways in which our transportation system may operate in the future, will
likely not significantly change total electricity demand from transport, which is the goal of this paper.
Our review of the literature can be found in sections V.A – V.D.
Seamless mobility systems, which integrate public transport with “last mile” taxi services, could shift
enough transportation away from individual LDVs to have a small impact, and we conduct a simple
calculation to bound the power implications of a concerted public policy push towards SMSs. The result
of this calculation (found in Section V.E – Seamless Mobility Systems) is a potential 2% drop in electricity
demand.
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
9 Electric_Transport_Draft _10_1_17
Wild Card factors
Finally, we consider “wildcard factors” – a handful of factors that will influence future LDV travel in
especially unpredictable ways. We look at (1) Road Infrastructure Costs, including AV-specific
infrastructure, and the manner in which LDV travelers will or will not pay for it; (2) telecommuting, e-
commerce, and other electronic substitutes for personal or business travel; and (3) redesign of urban
areas to reduce the need to travel.
The three “wild cards” we have surveyed have generally done a poor job of living up to their label. Of
the three, we have concluded that electronic travel substitutes are unlikely to result in VMT differences
not already captured in the range of outcomes in the three layers of modeling above. Our review also
indicates that urban design will, at most, add 2% on top of our existing scenarios. As urban redesign is
largely policy-driven, not an exogenous factor, our non-policy scenarios amount to a prediction that the
most likely outcomes exclude a significant policy shift that could, if adopted, reduce travel.
Charges for the use of infrastructure in a manner that affects driving is also a true wild card. It is far
beyond our ability to predict how the U.S. federal government and the states will cope with the
deterioration of existing roads and the need for infrastructure to service AVs. Even today, well before
the advent of AV-specific infrastructure, these questions push the U.S. Congress and many states to the
political breaking point. About all that can be said of this wild card is that it, too, presents almost
entirely downside risk to transport power demand. Today, no LDV pays anywhere near its full share of
the cost of roadway infrastructure; total infrastructure funding is far short of funding needs; and as yet
electric vehicles pay even less than gasoline cars.
We believe we can get a rough, order-of-magnitude range by examining two simple pricing scenarios: a
flat 2.2 2017 cents per mile charge and a larger 2.4 cents per mile ($.60/gal @ 25 mpg) escalating to
double its level in real terms by 2050.
Snapshot results of the effect on VMT in the year 2050 can be expressed in a 4x4 matrix, below:
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
10 Electric_Transport_Draft _10_1_17
Table 6: Road Pricing Reductions in VMT
The result is a 10% to 42% reduction in VMT in the year 2050 when the full effects of our four VMT fee
scenarios are applied. However, we use only the -.2 elasticity for our two base scenarios.
Results and Observations
The authors of the hundreds of pieces of research we have relied upon have each made dozens of
assumptions underlying their work. As we have compiled this research we have made dozens more.
Were we to catalog these comprehensively, we would end up with a huge list and an infeasibly large
number of possible scenarios and sensitivity runs that could be examined.
However, over the course of our research a handful of assumptions stand out as particularly important,
either because they describe an important fork in the development path for U.S. passenger transport or
because they have relatively strong and direct effects of LDV power use. A full summary of these
variables is included in chapter VII. In brief, they encompass: 1) projections for vehicle sales and
adoption, changes in VMT from EV price signals and AV technological improvements, gains in energy
efficiency from both EV and AV evolutions (EI), and potential road pricing and policy signals.
The table below summarizes LDV transport power demand from our calculations for the milestone years
between now and 2050. As the table shows, 2050 LDV power use is approximately 1140 TWh and 570
TWh, in the High Base and Policy Cases, respectively. As these cases are intended to approximate
upper and lower likely boundaries, the results are surprisingly close together. Whereas the earlier
literature surveys described in Chapter II of the report found upper and lower bounds differing by as
much as a factor of ten, our calculations suggest that the difference between our likely boundary cases
is only about 600 TWh, 15 percent of today’s power use. If our calculations have any value, we have a
pretty good idea of the power we’ll need thirty years from now so long as EVs take off on roughly the
high sales trajectory we forecast and no black swan events cause a serious rupture in American driving
habits.
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
11 Electric_Transport_Draft _10_1_17
Table 7: Results Summary
Electricity Consumption Summary
Case Year
Total Number of
EV in Service
Portion Stock
Electric
Total Number of
AV in Service
Fleet Average eVMT / Vehicle
Fleet Average
Efficiency Total TWh
(%) (per yr) (kWh/mile) (TWh)
High Base
2015 406,076 0.2% 0 7,179 0.32 1
2025 16,890,719 6.5% 0 9,087 0.34 53
2030 52,379,566 19.7% 3,182,833 10,290 0.35 187
2040 166,979,970 59.6% 65,615,683 13,420 0.33 742
2050 252,371,537 85.6% 180,263,265 16,927 0.27 1140
Policy Case
2015 406,076 0.2% 0 7,179 0.32 1
2025 17,086,996 6.6% 0 8,508 0.31 45
2030 52,378,548 19.7% 196,278 8,826 0.30 140
2040 166,928,240 59.6% 17,786,550 8,865 0.29 435
2050 251,932,162 85.5% 128,559,496 10,038 0.23 570
Section VII.C dives further into sensitivity analyses and scenario decomposition, and lead us to the same
overall conclusion. It is beyond both our means and expertise to provide anything approaching a
complete discussion of the implications of our findings for energy, transport, or climate policy. Instead,
we provide a small set of policy observations that speak mainly to the focus of our analysis, namely the
intersection of transport changes and the power industry. Beyond electrification of LDVs per se, the
policy approaches to reducing carbon seem to divide into these categories:
(A) shift drivers – and later, single occupants of AVs -- out of SOVs and into either pooled rides
or, much better, integrated multimodal on-demand mobility systems, via any number of
policy tools;
(B) encourage or require electric LDVs to become more efficient more quickly than otherwise,
much as CAFE and ZEV standards have forced ICE fleet efficiency gains; or
Base
High
Policy
Case
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
12 Electric_Transport_Draft _10_1_17
(C) Harvest the vehicle and system efficiency improvements theoretically offered by AVs as
soon as possible after they are introduced.
From the policy standpoint, the autonomous vehicle revolution is exceedingly complex. This is an area
where much more work is needed. We need much better data on the realistic changes we will need to
make to our road and communications infrastructure to accommodate AVs at each penetration level,
and how these changes can be staged so they need not be completely redone as the AV fleet grows.
We also need better data on how these vehicles will co-exist with conventionally-driven cars and trucks
and how efficiency and safety improvements can be accelerated in the presence of mixed fleets.
Finally, there is almost no data on how much the infrastructure changes for AVs will cost, much less on
how we will finance them.
With the possible exception of the latter, enormous amounts of research are now underway. When we
have some of these answers we will have the ability to make somewhat better estimates of the impacts
of autonomous vehicles on future power demand.
Executive Summary Please see http://www.bu.edu/ise/what-we-are-working-on/ for full paper with appendices.
13 Electric_Transport_Draft _10_1_17